- After seeing how merely quantizing a model can make it go berserk, I have very little confidence I can fine tune an LLM and expect similar performance benchma

- After seeing how merely quantizing a model can make it go berserk, I have very little confidence I can fine tune an LLM and expect similar performance benchmarks. - A RAG-empowered LLM can tell you where the knowledge used to answer a question came from.